Active monitoring of border network fabrics

Patent metadata

Field	Value
Publication number	US-10243785-B1
Application number	US-201615182133-A
Country	US
Kind code	B1
Filing date	Jun 14, 2016
Priority date	Jun 14, 2016
Publication date	Mar 26, 2019
Grant date	Mar 26, 2019

Abstract

Technologies are provided for active monitoring of network fabrics, such as border network fabrics. For example, test packets can be generated and sent to destination devices within the network fabrics. The test packets can have destination ports that have been identified as closed destination ports. The success of the test packets can be determined based on whether ICMP error packet responses are received. If a network problem is detected based on results of the test packets, then traceroute packets can be used to identify the location of the network problem.

First claim

What is claimed is:

1. A method, implemented by a computing system, for monitoring border network fabrics, the method comprising:
generating, by a network monitoring agent located outside a border network fabric, user datagram protocol (UDP) test packets, wherein the UDP test packets:
are addressed to a destination device in the border network fabric and
have a destination port that has been identified as a closed destination port;

sending, by the network monitoring agent, the UDP test packets for delivery to the destination device;
in response to a first UDP test packet, receiving an Internet control message protocol (ICMP) error packet that indicates the first UDP test packet was successfully received by the destination device, wherein an ICMP error packet is one of a destination port unreachable ICMP error packet or a source quench ICMP error packet;
in response to a second UDP test packet sent after the first UDP test packet, receiving a result that is other than an ICMP error packet, the result indicating that the second UDP test packet was not successful; and
responsive to determining that the second UDP test packet was not successful, sending traceroute packets to identify a source of a network error within the border network fabric.

2. The method of claim 1 wherein the result of the second UDP test packet that is other than an ICMP error packet response comprises:
a response packet other than an ICMP error packet; and
a lack of a response packet received during a timeout period.

3. The method of claim 1 wherein the border network fabric comprises a multi-stage network with a plurality of stages, wherein each stage of the multi-stage network comprises a plurality of switches or routers.

4. The method of claim 1 wherein the border network fabric comprises:
a transit layer of network devices, the transit layer providing transit and peering to external networks;
a distribution layer of network devices, the distribution layer managing traffic flow for multiple data center areas; and
an aggregation layer of network devices, the aggregation layer aggregating traffic for each of the data center areas.

5. The method of claim 4 wherein the destination device of the UDP test packets is a network device in the transit layer.

6. A computing device comprising:
a processor; and
a network interface;
wherein the computing device is configured to perform operations for monitoring network fabrics, the operations comprising:
for each of a plurality of test packets:
generating the test packet, wherein the test packet:
is addressed to a destination device in a network fabric; and
has a destination port that is a closed destination port;

sending, via the network interface, the test packet for delivery to the destination device in the network fabric; and
determining if the test packet was successful based on whether an Internet control message protocol (ICMP) error packet response is received;
wherein the test packet is successful if the ICMP error packet is received in response to the test packet;
wherein the test packet is unsuccessful if:
a response to the test packet is received indicating a response other than the ICMP error packet, which indicates a network problem within the network fabric; or
a response to the test packet is not received within a timeout period, which indicates a network problem within the network fabric.

7. The computing device of claim 6 , the operations further comprising:
calculating a rate of unsuccessful test packets over a period of time; and
when the rate of unsuccessful test packets is over a threshold value, sending traceroute packets to identify a location of a network problem within the network fabric.

8. The computing device of claim 7 wherein the traceroute packets are configured to use header parameters and overall packet size of the test packets.

9. The computing device of claim 6 wherein the computing device is located external to the network fabric.

10. The computing device of claim 6 wherein the destination device is any computing device within the network fabric with an Internet Protocol (IP) network stack.

11. The computing device of claim 6 wherein the network fabric is a border network fabric that has multiple layers.

12. The computing device of claim 11 wherein the multiple layers comprise a transit layer, and wherein the destination device to which the test packet is addressed is in the transit layer.

13. The computing device of claim 6 wherein the test packet is one of a user datagram protocol (UDP) ping packet or a transmission control protocol (TCP) ping packet.

14. A method comprising:
for each of a plurality of test packets:
generating, by a monitoring agent, the test packet, wherein the test packet is:
addressed to a respective destination device in a border network fabric; and
has a respective destination port that is a closed destination port;

sending, by the monitoring agent, the test packet for delivery to the destination device in the border network fabric;
determining if the test packet was successful based on whether a response received to the test packet is an Internet control message protocol (ICMP) error packet; and

determining whether there is a network problem within the border network fabric based on results of the sending of the test packets.

15. The method of claim 14 wherein the test packet is one of a user datagram protocol (UDP) ping packet or a transmission control protocol (TCP) ping packet.

16. The method of claim 14 wherein the plurality of test packets are configured to vary network paths taken within the border network fabric by varying one or both of the destination port among a plurality of closed destination ports or a source port among a plurality of source ports.

17. The method of claim 14 wherein the network monitoring agent is a software agent running on a host external to the border network fabric.

18. The method of claim 14 wherein determining whether there is a network problem within the border network fabric comprises:
calculating a rate of unsuccessful test packets over a period of time; and
when the rate of unsuccessful test packets is over a threshold value, determining that there is a network problem.

19. The method of claim 14 wherein when a network problem is determined, sending traceroute packets to identify a location of the network problem within the border network fabric.

Assignees

• Amazon Tech Inc

Inventors

• Corrigan John
• Mccabe Karl Andre
• Cavanagh Eoin Francis

Classifications

• H04L43/0823

  Errors, e.g. transmission errors · CPC title

• H04L43/50

  Testing arrangements · CPC title

• H04L43/10

  Active monitoring, e.g. heartbeat, ping or trace-route · CPC title

• H04L41/0677 Primary

  Localisation of faults · CPC title